On this page
In 2025, 76% of surveyed organizations said less than half of their CRM data was accurate and complete, according to WhispersLab's account of Validity's survey of 602 CRM users and stakeholders in the United States, United Kingdom, and Australia. That's a different problem from missing email addresses. Teams can have plenty of B2B contact data and still lack reliable evidence about whom they're contacting.
A useful contact record isn't a permanent fact. It's a dated hypothesis with a verification state: this person appears to hold this role at this company, and this contact method passed these checks at this time. Build around that distinction, and enrichment becomes a controlled workflow rather than a larger pile of names.
Table of Contents
- What a B2B Contact Record Contains
- Keep the person and account separate
- Where B2B Contact Data Comes From
- Choose evidence, not just a provider
- Why Contact Records Go Stale
- A job move breaks more than an address
- How to Verify Emails and Phones Before Use
- Keep identity confidence separate from sendability
- Check phones on their own terms
- Enriching Contacts and Avoiding Costly Mistakes
- Build the workflow around acceptance gates
- Fix the mistakes that create downstream work
- Give automated outreach an audit trail
- Test Enrichment on Your Own List Before You Commit
- Run a controlled comparison this week
What a B2B Contact Record Contains
| Field | What it tells you | Usual failure | How to check |
|---|---|---|---|
| Full name | Who to address | Nicknames and name changes | Match against a current profile |
| Job title | Role and seniority | Changes with promotions or moves | Compare with a recent profile |
| Current employer | Which account owns the record | Job changes | Confirm the company still employs the person |
| Work email | A way to reach the person in writing | Bounces, closed domains, role addresses | Syntax, MX, SMTP and risk checks |
| Direct phone | A way to reach the person by call | Wrong numbers, unreachable lines | Reachability check |
| Location | Region for timing and consent rules | Remote work and relocations | Compare profile and company data |
These fields describe different things, so they shouldn't share a blanket accuracy label. A current name doesn't establish a current employer. A reachable phone line doesn't establish that the intended person answers it.
Presence is not usability. A name and email in a database are stored values. An email with a verification result and timestamp gives your sending system evidence for a contact decision, though it still doesn't guarantee delivery.
Keep the person and account separate
The person record usually connects to the company record through an employer relationship and a normalized domain. That connection lets operations teams assign ownership, apply account exclusions, and route leads without matching company names manually.
Current employer is the relationship to distrust first. When someone moves, the person may remain identifiable while the account association becomes wrong. Don't merge records solely because an old email domain matches an existing account.
Company-level fields sit beside this relationship: domain, headcount, industry, and company location describe the account, not the individual. A headquarters address doesn't tell you where a remote employee works, and company size doesn't establish someone's purchasing authority.
Schema rule: Keep identity, employment, contactability, and account attributes separate. They answer different questions and require different evidence.
For outreach, request email and phone with their check results. For targeting, prioritize employer and title, then map titles to your own role taxonomy rather than treating every director as the same buyer.
Use location for working-hour scheduling and as an input to jurisdiction review. Don't infer consent or applicable outreach rules from a country field alone. The practical minimum is a record that explains who the person is, why they fit, and which channel is usable.
Where B2B Contact Data Comes From
| Source type | Strength | Weakness | Typical use |
|---|---|---|---|
| Company websites | Official titles, contact pages and domains | Often lack named people and personal emails | Resolving the business and its domain |
| Public professional profiles | Current titles and employers | Self-reported and can be outdated | Confirming role and company |
| Licensed third-party datasets | Broad coverage at scale | Coverage differs by market and sources can disagree | Filling emails and phone numbers |
| Multi-provider waterfall | Additional matches when an earlier source misses | Conflicting answers need a rule | Enrichment pipelines |
| Own CRM and forms | First-party consent and interaction history | Limited coverage and often stale | Deduplication and reactivation |
Official company sources are useful for identifying the business. They're less dependable as a complete directory of employees, especially when the website only lists executives or a shared contact address.
People-level sources solve a different problem. Professional profiles can connect a name to a role and employer, but the person controls when the profile changes. A page retrieved today may still describe employment that ended earlier.
Choose evidence, not just a provider
A professional-profile integration needs identity matching, not merely a successful response. When designing a LinkedIn data API workflow, retain the returned profile URL so you can distinguish the intended person from someone with the same name.
A waterfall queries providers in sequence until a result meets your acceptance criteria. This can recover matches an earlier provider missed, but stopping at the first nonempty response trades precision for coverage.
Set a conflict rule before connecting the waterfall to your CRM. Prefer the most recently verified value when the checks are comparable; otherwise, retain both candidates and flag the field for review. A recent retrieval timestamp shouldn't beat older evidence merely because the retrieval happened later.
- Preserve provenance: Store the provider, source reference, retrieval time, and any source update time separately.
- Handle disagreement explicitly: Mark conflicting employment or contact values rather than picking one.
- Separate regional coverage: Evaluate North America, EMEA, and APAC independently instead of assuming a source performs uniformly.
First-party forms often give you the clearest history of what someone submitted and which notice or permission accompanied it. That doesn't make every form submission permission for every campaign. Preserve the actual notice, purpose, and preference history so downstream systems can decide whether a proposed use fits.
The best source mix isn't necessarily the broadest. It's the mix whose evidence you can explain and whose gaps your workflow can handle without guessing.
Why Contact Records Go Stale
U.S. Bureau of Labor Statistics tenure figures cited in Forage's analysis of B2B data decay show that 20.6% of wage-and-salary workers had been with their current employer for one year or less in January 2026, while median employer tenure was 4.1 years. These figures measure labor mobility, not CRM decay directly.
The same analysis describes a commonly repeated 2.1% monthly accuracy-loss estimate, compounding to about 22.5% over twelve months, but notes that the original MarketingSherpa methodology isn't publicly documented. Treat it as a planning assumption, not a measured rate for your database.
| Figure | Value | Source status | How to read it |
|---|---|---|---|
| Workers with current employer one year or less, January 2026 | 20.6% | BLS tenure data via independent analysis | Labor mobility, not CRM decay |
| Median employer tenure | 4.1 years | BLS tenure data via independent analysis | Employment context, not a record expiry date |
| Monthly accuracy-loss benchmark | 2.1% | Commonly repeated; MarketingSherpa method undocumented | Planning estimate only |
| Compounded over twelve months | About 22.5% | Derived from the monthly benchmark | Estimate, not a universal measurement |

A job move breaks more than an address
Say your CRM lists a marketing director at a software company. The person joins another employer, but your automation keeps the original record active.
A sequence may bounce at the old address, reach a replacement through a reassigned mailbox, or continue with messaging based on the wrong role. Meanwhile, the account owner believes the original company still has a known marketing contact.
Employer, title, and work email are especially exposed to employment changes. Promotions create a subtler failure: the mailbox may work while the targeting assumption becomes wrong.
Don't assign the whole record an expiry date based on a generic decay percentage. Schedule checks by field and use: employment before role-sensitive prospecting, contact methods before channel activation, and account attributes when segmentation depends on them.
A contact record should say what you believe, when you checked it, and what would make you stop using it.
For global GTM operations, refresh frequency and employment verification belong in the pipeline design. A periodic spreadsheet cleanup can't protect an automated system that continues acting on stale values between cleanups.
How to Verify Emails and Phones Before Use
| Check | What it catches | What it cannot prove |
|---|---|---|
| Syntax | Typos and malformed addresses | Whether the mailbox exists |
| Domain and MX | Domains that cannot receive mail | Whether the person still works there |
| SMTP acceptance | Many closed or unknown mailboxes | That the message reaches the inbox |
| Role and disposable screening | Shared or throwaway addresses | Identity or intent |
| Historical bounce outcomes | Addresses that failed before | Current status without a recheck |
| Timestamp and method | When and how the address was verified | Whether it still works today |
A 2026 analysis of email-verification outcomes reports that ZeroBounce checked more than 11 billion email addresses during January–December 2025 and found 23% invalid. Those results span industries and company sizes; they aren't a failure-rate forecast for your particular B2B list.
The operational lesson is straightforward: a database count tells you little about how many addresses will pass checks when you need them. Verification belongs between acquisition and activation, not just inside a provider's product description.

Keep identity confidence separate from sendability
SMTP acceptance isn't proof of inbox placement. A receiving system can accept a message and later classify it as spam, and an accepted mailbox may not be monitored. Validity's Email Deliverability Benchmark supports treating list hygiene and sender controls as complementary requirements.
Store verificationState, verifiedAt, verificationMethod, and confidence alongside the address. Keep raw provider evidence intact, then calculate a separate sendability decision from those results, suppression status, and your campaign policy.
This separation lets you retain a correctly identified person while blocking a risky address. It also avoids the opposite mistake: assuming that a technically acceptable mailbox belongs to the intended buyer. A practical email verification workflow needs both technical checks and explicit handling of uncertain outcomes.
Check phones on their own terms
Normalize phone numbers with country context before checking them. Record whether the result indicates a reachable line, what line type was returned, and whether the number is direct or a switchboard.
Reachability doesn't prove ownership. A connected line may belong to reception, a former employee, or someone unrelated. Feed wrong-person outcomes back into the record and suppress the person-number pairing rather than marking the entire person invalid.
Use separate email and phone states. A failed email shouldn't automatically block a confirmed phone number, while an unanswered call shouldn't be recorded as proof that the line is disconnected.
Sender authentication remains a separate control. SPF authorizes sending infrastructure, DKIM signs messages, and DMARC connects authentication to domain-alignment policy; none establishes that a prospect exists. Recheck addresses before high-volume campaigns rather than relying solely on import-time verification.
Enriching Contacts and Avoiding Costly Mistakes
A reliable sales data enrichment workflow starts with entity resolution and ends with controlled writes. The dangerous shortcut is allowing a successful lookup to trigger an unconditional CRM update.
Build the workflow around acceptance gates
- Start with the full legal company name. Send the legal name rather than a trading short form to reduce ambiguous matches. Include known location or domain evidence where available, and retain the original input for review. The common slip is accepting a similarly named company because its result looks complete.
- Resolve the business website first. Confirm the returned website belongs to the intended entity before searching for people. Store the normalized domain against the company record, retaining the original website URL as evidence. Don't guess a domain from a brand name and then use that guess as proof for every later lookup.
- Search for a named contact only when the inputs support it. A website and a person's name give the finder a grounded lookup. Single-location businesses frequently have neither a named contact nor a personal work address, so an unresolved person can be an honest result. A generic mailbox shouldn't be relabeled as a named decision-maker to make the enrichment appear successful.
- Verify email and phone independently. Capture the outcome, time, method, and any unresolved risk for each channel. Route uncertain results into a review or recheck queue rather than forcing every response into valid or invalid. The common slip is using email validation as a blanket approval for all contact fields.
- Compare employment before updating the CRM. Check the returned employer and title against the existing record. A mismatch could indicate a job move, an incorrect profile match, or an outdated internal record. Preserve the conflicting evidence and resolve it before changing account ownership or launching a sequence.
- Write accepted fields, not entire responses. Update only fields that meet your verification policy; retain other candidates as unverified evidence. Keep suppression flags outside ordinary enrichment overwrite rules. Deduplicate before upload so repeated imports don't split activity history or recreate an excluded contact.

Use idempotent writes so retrying a job doesn't create duplicate people or accounts. Store the lookup input and accepted output together, allowing an operator to reconstruct why a field changed without rerunning the request.
Fix the mistakes that create downstream work
Validity's survey finding about incomplete and inaccurate CRM records points to a practical problem: teams need acceptance rules, not merely more incoming fields. These mistakes turn enrichment spending into manual cleanup:
- Treating found as contactable: Require an approved channel state before activation. Otherwise, salespeople spend time on wrong recipients and failed contact attempts.
- Judging a database by size: Compare the usable share of returned records, not catalog volume. Broad coverage can still leave you paying to review unusable candidates.
- Never comparing sources: Run the same inputs through an alternative source using identical acceptance rules. Without that comparison, you can't distinguish a genuine coverage gap from a provider-specific weakness.
- Skipping suppression propagation: Apply opt-outs across CRM, sequencing, enrichment, and agent tools. Recreating an opted-out contact can lead to unwanted repeat outreach and complaint handling.
- Using a global accuracy score: Separate email and phone results by region and field. Targetwise's contact-data benchmarking methodology argues for layered mobile-accuracy reporting rather than a single global figure.
The last mistake matters economically. A wrong phone number consumes calling capacity; an incorrect employer can route the prospect to the wrong representative. Measure those outcomes separately instead of hiding them inside a blended match rate.
Give automated outreach an audit trail
An agent needs more than access to a contact endpoint. It needs a policy that blocks action when identity is uncertain, suppression applies, or the proposed message lacks relevance.
For each attempted action, retain a decision receipt containing the selected person, source references, verification evidence, suppression result, purpose, and allow-or-block reason. Include the fields the agent used, not just its final explanation. An explanation without underlying evidence isn't an audit trail.
Governance concerns described in Apollo's discussion of B2B contact databases support keeping provenance and lawful-basis metadata alongside operational data. Where legitimate interests is the applicable basis, assess purpose, necessity, and balancing rather than treating professional relevance as automatic permission.
Outreach gate: Contact someone only when the role fits the offer, the evidence meets your freshness policy, suppression checks pass, and the contact is justified under the applicable rules.
Legitimate interests isn't a universal exemption from channel-specific requirements. Repeat outreach should have a reason, such as relevant new information, rather than continue because a sequence still has steps remaining.
Test Enrichment on Your Own List Before You Commit
Test providers on your records before committing to their coverage claims. Use the same inputs, acceptance criteria, and verification process so you're comparing usable outcomes rather than different definitions of a match.
Run a controlled comparison this week
- Select 200 contacts whose emails are already known to be valid. Keep the known addresses as a reference set, but remove them from the lookup input. Include contacts representative of the markets and roles you target.
- Resolve websites, then run the lookup. Send names and company websites through
email_finder, followed byemail_verifier. RichAPI is one option to include in this comparison, using the same acceptance rules as every other source. - Validate the returned identity. Send the returned profile URL to Profile Enrichment API and compare
currentCompanyandcurrentTitlewith your own record. Count a mismatch as a failed match for the benchmark, then investigate whether the provider or your reference record needs correction. - Calculate cost per usable contact. Divide total lookup and verification cost by the number of results that pass your criteria. Include review costs separately if operators must resolve conflicts.
- Record the complete outcome. Track hit rate, verified share, mismatches, and elapsed time for each source. Separate a plausible returned address from a verified, correctly matched contact.
Choose the source with the best verified cost per usable contact, not the largest pile of returned values.
A known-valid reference set tests recovery, not discovery across your entire addressable market. Keep that limitation in the decision, and don't treat every different returned address as wrong if an independently verified alternative exists.
Keep sources that meet your identity, reachability, and cost requirements. Drop those with poor verified share or excessive mismatches, schedule refreshes for accepted records, and preserve the evidence behind each decision.
RichAPI offers a B2B data enrichment API and MCP server for contact, company, social, and signal data from licensed providers. Visit RichAPI to build a test using company resolution, email finding, verification, and profile enrichment, then compare the usable results against your own records before expanding the workflow.
